News
Experts’ dismissal of Tesla’s Full Self Driving push proves Elon Musk is still not taken seriously
Despite delays, Elon Musk remains optimistic that Tesla’s efforts to achieve Full Self Driving will bear fruit soon. This was highlighted in a statement during his appearance at the recently held 2020 World Artificial Intelligence Conference in China, where he noted that Tesla is closing in on “Level 5” functionalities. Experts in the field of automation beg to differ.
In a recent statement to Automotive News, Sam Abuelsamid, principal analyst at Guidehouse Insights (formerly Navigant Research) and Missy Cummings, director of the Humans and Autonomy Laboratory at Duke University, openly dismissed Tesla’s efforts to achieve full self driving. In a scathing rebuke, Abuelsamid described Musk’s targets as absurd. He also expressed his pessimism about Tesla’s approach to automation, which focuses on vision and artificial intelligence.
“The cars they are building will never be Level 5, period. It’s nonsense. He needs to shut up until he can deliver something. The premise of making highly automated systems on cameras alone is fundamentally flawed. Their approach to software doing end-to-end AI systems is almost certainly not going to work. I don’t believe it can work. AI is too brittle,” he said.
It should be noted that Abuelsamid recently published a study ranking companies that are currently pursuing autonomy. Similar to studies conducted by Guidehouse Insights when it was still operating as Navigant Research, Abuelsamid ranked Tesla dead last, even behind startups like Navya, which have limited real world driving data.
Abuelsamid’s sentiments were echoed by Missy Cummings, who also argued that there is no way that Tesla could achieve Level 5 autonomy with its vehicles. In a statement to KCBS Radio, Cummings noted that Elon Musk likely does not understand what Level 5 Automation really is. This is an interesting comment from the director of the Humans and Autonomy Laboratory at Duke University, considering that Waymo CEO John Krafcik noted last year that ultimately, “Level 5 is a bit of a myth.”
“There is no way, shape, or form the car is going to be Level 5. He doesn’t really understand what Level 5 is. I think what he means is Level 4, and he’s not even going to get Level 4. The perception systems don’t work well in weather conditions, with long shadows. We know the Tesla perception system, and the news is it’s a really bad system,” she remarked.
While there is some merit in criticizing the Tesla CEO for missing his targets with regards to the release of the electric car maker’s full self driving system, it seems far too careless to simply dismiss all the work that Tesla has accomplished over the years either. The company, after all, is gathering real world driving data at an unprecedented scale that’s unrivaled by companies like Waymo, and that matters much when it comes to training neural networks, as emphasized by ARK Invest CEO Cathie Wood. Tesla’s vision based approach to autonomy has also gained support from notable names such as George Hotz, who believes that Tesla would eventually become the Apple of self driving cars.
It is unfortunate, but based on the dismissive and almost aggressive rebuke of Tesla’s efforts from the experts tapped by Automotive News, it appears there is still a general tendency to completely dismiss Elon Musk and his initiatives. Interestingly enough, these statements are pretty familiar to those who have followed the SpaceX and Tesla story over the years. It may be difficult to recall at this point, but there was a time when it was believed that landing the Falcon 9 first stage on land and an autonomous barge on the sea is impossible. There was also a time when the Model X was considered unbuildable.
Fortunately, Tesla and its CEO are a stubborn bunch, and it takes a lot more than skepticism to discourage the company. This is especially notable considering that Tesla is currently looking to roll out a massive rewrite of its Autopilot suite, which should take the company even closer to full self driving. Perhaps this time around, Elon Musk’s statements about Level 5 Autonomy may actually be pretty accurate, at least in terms of its features and timeframe.
“I’m extremely confident that level five or essentially complete autonomy will happen, and I think, will happen very quickly,” said Musk, answering another question. I think at Tesla, I feel like we are very close to level five autonomy. I think—I remain confident that we will have the basic functionality for level five autonomy complete this year,” Musk said at the 2020 WAIC.
News
Tesla’s two defunct flagship models are getting a big upgrade
Tesla’s two recently-defunct flagship models, the Model S and Model X, are getting a big upgrade, according to the company’s Head of AI, Ashok Elluswamy.
Older Hardware 3 Model S and Model X vehicles have been the last major holdouts in Tesla’s Full Self-Driving v14 Lite rollout, and that wait now appears to be ending.
Tesla brings closure to flagship ‘sentimental’ models, Musk confirms
At Tesla’s Cybercab launch, AI chief Ashok Elluswamy told Ryan McCaffrey that he thought the S and X build “was supposed to go out last week.” Evidently, Elluswamy expects the suite to be rolled out to those HW3 Model S and Model X very soon:
For my @Tesla friends – and specifically Model S & X owners with HW3 who are waiting on FSD v14 Lite – I spoke to @aelluswamy at the Cybercab launch, & when I asked for an ETA on v14 Lite for S/X owners, he said, “Oh, I thought it was supposed to go out last week.”
So: soon! 🙌
— Ryan McCaffrey (@DMC_Ryan) September 7, 2026
Those cars are not the current Model S and Model X, which already ship with Hardware 4. They are the pre-refresh flagships built around Tesla’s older Autopilot computer, often called HW3 or AI3.
Tesla stopped putting that computer in new vehicles years ago, which is why owners treat these S and X cars as a closed generation. Model 3 and Model Y vehicles on the same computer began receiving v14 Lite in late June 2026 and saw a wider North American expansion in July. South Korea followed as an early international market. The S and X versions of the same software never joined that wave.
v14 Lite is Tesla’s way of squeezing the current v14 driving stack onto hardware that cannot run the full AI 4 model. The company describes the process as distillation: behaviors learned on the newer computer, including reinforcement learning and offline models, are compressed so the older chip and cameras can use them as a guide.
Early descriptions put the distilled network at roughly 15 percent of the original size. The result is still supervised Level 2 driving. Tesla has been clear that HW3 cannot support unsupervised Full Self-Driving or robotaxi operation because of memory and bandwidth limits.
The feature list is what made the wait so frustrating for S and X owners, as plenty of new features are to be shipped with it.
Official notes for the first Lite build, firmware 2026.20.5.1, added parking, unparking, and reversing; arrival options for a parking lot, street, driveway, or curbside; speed profiles that stay available at all times; and start-from-park engagement. Tesla also claimed better handling of merges, forks, pedestrians, traffic lights, and cut-ins, plus fewer false slowdowns and smoother lane centering.
A mid-July follow-on build, 2026.20.6.10, added more of the Hardware 4 interface, including a standalone Self-Driving app and the ability to start a trip from Park without a brake-pedal confirmation.
Elluswamy called that version the one “likely going to wide release.”
That wide release already reached most other HW3 cars in the United States and Canada. International timing still depends on regional validation and regulatory approval. For S and X owners, the remaining work appears to be model-specific validation rather than a new software stack.
There is no official Tesla changelog or build number for those two models yet, only Elluswamy’s offhand timeline. Some HW3 drivers who already have Lite report large gains over v12.6; others have described new indecision or phantom braking. The next test will be whether the same software lands cleanly on the older flagships that have waited the longest.
Cybertruck
This tiny Tesla Cybertruck adjustment has big advantages
Yesterday, we reported on Tesla Cybertruck getting some major adjustments from a manufacturing standpoint in an effort to make the all-electric pickup more cost-effective, more reliable, more serviceable, and more easily produced.
Tesla Cybertruck engineer reveals new changes in ‘constantly evolving’ pickup
One of those changes was the addition of a self-reinforcing polypropylene aero shield that sits underneath the truck. Previously, Tesla utilized aluminum for this, but the self-reinforcing polypropylene was more durable while also being cheaper and lighter.
Tesla has revealed another small change it made to the Cybertruck, and it has to do with the side repeater cameras.
Tesla does not wait for a new model year to improve its vehicles. On September 8, Cybertruck lead engineer Wes Morrill posted side-by-side photos of an updated side repeater camera housing now rolling off the line at Gigafactory Texas.
The triangular camera pod mounted on the front fender looks almost identical at first glance. A closer look reveals a revised contour that uses the air already flowing around the truck to keep the lens clearer in rain and road spray.
The side repeater camera was updated – the version on the left is the newer part which uses passive geometry to create airflow disturbance that better keep water off the lens while driving. No cost penalty, just pure vision improvement. pic.twitter.com/wAbXtcL1Jf
— Wes (@wmorrill3) September 8, 2026
The side repeater cameras sit in an exposed position on the Cybertruck’s angular stainless-steel body.
In wet weather, they readily collect water droplets that can degrade the image Autopilot and Full Self-Driving use for lane changes and blind-spot monitoring. Early production trucks sometimes left owners wiping lenses by hand or accepting temporary restrictions on driver-assistance features.
Tesla has added washers to cameras on certain other models and on Cybercab prototypes, but those active systems add cost, complexity, and extra potential leak points.

The new housing solves the problem with passive geometry. Subtle changes in the surround create localized airflow disturbances as the vehicle moves. Those eddies physically push water droplets away from the optical surface. Morrill called the result “pure vision improvement” achieved at “no cost penalty.” Once the production mold is updated, every subsequent part costs the same as the original.
The advantages compound quickly. Clearer cameras in rain improve the reliability of driver-assistance features precisely when they are needed most. The design consumes no extra energy and introduces no new failure modes.
New Cybertrucks built after the tooling changeover receive the updated part automatically. Some owners of trucks delivered as late as June 2026 have already confirmed they received the revised housing. Retrofit questions have appeared in replies, and the cameras appear electrically compatible, though Tesla has not announced an official service program.
A few millimeters of reshaped housing will not make headlines the way a new battery pack does, but these changes are incremental and increase the Cybertruck’s effectiveness as a vehicle over time.
This improvement illustrates how Tesla continues to refine the Cybertruck after volume production began. Better wet-weather vision, zero added cost, and no extra hardware add up to a meaningful gain in everyday usability and safety.
News
Tesla is rolling out a new FSD version with a massive safety addition
Tesla is rolling out a new version of its Full Self-Driving suite to some owners that comes with the massive addition of a safety feature.
Tesla is rolling out Automatic Collision Evasion with the 2026.27.6 Software Update, which started rolling out to some vehicles last night. We received the update, along with Full Self-Driving v14.3.9, as well as v14.2 Lite, which has identical release notes as the previous version and seems to have some refinements and improvements in behavior and performance.
🚨 Tesla FSD v14.3.9 is rolling out as well as 2026.27.6 which includes Automatic Collision Evasion
Brand new safety features from a software update. My Tesla gets better everytime I get one of these pic.twitter.com/ctrMiQXWhM
— TESLARATI (@Teslarati) September 8, 2026
However, most of the attention has fallen on the Automatic Collision Evasion feature, which we covered in an article last week.
The function will activate Full Self-Driving to “try to keep your vehicle safe and then continue driving. It can engage in the following situations while you are driving manually:
- Scenario 1: A frontal collision is imminent and braking alone may not avoid it.
- Scenario 2: Your vehicle detects that you are not sufficiently attentive to the road (for example, reaching toward the back seat), or that Full Self-Driving (Supervised) may have been unintentionally disengaged.”
Essentially, FSD will take over when the vehicle determines you are not paying sufficient attention or are heading toward a potential collision. The addition of this feature is incredibly useful as distracted driving is a major issue in today’s world.
Along with the new safety feature is Tesla FSD v14.3.9, which has no additional release notes compared to the previous version, but in my first drives, my first impression is that operation is great, and parking is still sort of a pain point.
Just took a 15-mile round trip to the gym and back
Pretty on par with what FSD is nowadays – really good. Not enough time to see what’s good and what’s bad, but these first rides on any version feel shockingly good. They’re all pretty identical https://t.co/WN1qTg4bhE
— TESLARATI (@Teslarati) September 9, 2026
Additionally, Tesla v14.2. Lite has arrived. A great review of that is available here:
— Zack (@BLKMDL3) September 9, 2026
The addition of an Automatic Collision Evasion feature is similar to that of other collision avoidance systems that are used by companies like Hyundai, Kia, and Genesis. These programs typically utilize radar and camera sensors to apply emergency brakes autonomously, though evasive steering in a manual driving mode is pioneered primarily by Tesla’s newest addition.